The foreseen trigger rate of 10 
events/sec allows
10 
sec readout time per event. In addition,
with the beam specifications proposed (5 
10 
ptc/sec
spread over a 1 cm 
area), we expect a rather low occupancy.
The time needed to readout a full chip (according to the
RD20 design) is of 4, 8 or 16 sec with the clock
of 40, 20 or 10 MHz respectively.
Consequently 40 MHz and 20 MHz readout frequencies
will not cause any dead time. But using 10 MHz clock
would make us loose all features of the RD20 chip
and cause 16 sec dead time.
A possible solution would be to use 20 MHz clock
and 100 nsec peaking time. The use of a deconvolution
algorithm would guarantee a
time resolution of 
50 nsec [29].
In this situation we
would have 2-3 beam particles recorded
per event and a double interaction recorded in 5%
of the cases.